The cyanobacteria Microcystis spp. and Raphidiopsis raciborskii (basionym Cylindrospermopsis raciborskii) are the most problematic harmful freshwater species worldwide. Laboratory culture studies have been conducted by researchers globally to identify the optimal environmental conditions for growth of these species. However, these studies mostly focused on one or two strains, limiting understanding of withinspecies strain variation, and how this variation contributes to species competition under different environmental conditions. There is also limited knowledge of whether the global laboratory data are sufficient to confidently predict growth patterns of Microcystis and R. raciborskii in situ. Hence, this thesis aimed to determine the withinspecies strain variation in growth, and use this information to study how this variation affects species population dynamics. The laboratory studies I undertook showed that intraspecific variation was greater than interspecific variation based on measurements of growth responses of four M. aeruginosa and eight R. raciborskii strains (isolated from two adjacent Australian reservoirs) individually to a range of light intensity and temperature conditions. There was greater variation between strains of R. raciborskii than M. aeruginosa in terms of growth rate, light attenuation coefficient (i.e., self-shading) and cell volume, highlighting the extent of variation in strain responses to environmental conditions. In comparison, M. aeruginosa always grew to a higher cell concentration than R. raciborskii under the same conditions, indicating the capacity for greater dominance of M. aeruginosa over R. raciborskii worldwide. To identify how the strain variability affects population dynamics between M. aeruginosa and R. raciborskii, pairwise competition between the 12 strains was simulated using a deterministic model, parameterized with my laboratory measurements of growth and light attenuation for all strains, and run at two temperatures and two light intensities under well-mixed conditions. In total, 17,000 runs were simulated for each pair using a phytoplankton dynamics model incorporated with a Monte Carlo statistical approach, which was to propagate the variability and uncertainty of parameters in the deterministic model. The model outputs showed that cyanobacterial competition outcomes were highly variable, depending on the strains present, as well as the light and temperature conditions. Unlike “competitive exclusion” predicted by previous models, i.e., the most competitive species dominates and drives the other species to extinction, my study found no absolute ‘winner’ under all conditions. There were always strains predicted to co-exist with the dominant strain. This strain co-existence led to uncertainty in the prediction of species competition outcomes, which was due to the substantial variability in growth responses within and between strains. Overall, this study highlights that strain variability may substantially reduce our confidence in predicting phytoplankton competition, as deterministic models are typically based on only one set of parameters for each species or strain. Improving the ability to predict growth, dominance and dynamics of harmful cyanobacteria is critical to manage and control cyanoHABs. Based on my collation of studies over the last 30 years, Microcystis spp. and R. raciborskii dominate in 78% and 17%, respectively, of freshwater bodies across the globe that are affected by cyanobacterial blooms. I tested the hypothesis that light and temperature are the key drivers of observed differences in dominance by these cyanobacteria on a global scale. I synthesized growth responses of M. aeruginosa (as a proxy for Microcystis spp.) and R. raciborskii from 20 and 16 culture studies, respectively, to predict growth rates (d-1) as a function of light and temperature, including uncertainty in the growth rate predictions of each species. Variation in the dominance of the two species across latitudes in summer was predicted from a population dynamics model. The growth rate of R. raciborskii was predicted to exceed M. aeruginosa at temperatures ≳ 25oC and light intensities ≳ 150 μmol photons m-2 s-1. The population dynamics model predicted that M. aeruginosa would outcompete R. raciborskii at high latitudes of the temperate zones, but be outcompeted by R. raciborskii at tropical and subtropical zones, with coexistence in the lower temperate zones. Field observations, however, indicate widespread dominance of Microcystis over R. raciborskii irrespective of climatic zones. This is unlikely to be resolved by repeated laboratory studies that test environmental responses in well-mixed cultures with fixed environmental conditions. Instead, the key physiological attributes of colony formation, buoyancy and photoadaptation, and strainlevel variation are needed to develop process-based predictive models for cyanobacteria. This thesis has provided important new findings on the contribution of strains to cyanobacterial species variation and the effect of physical conditions to cyanobacterial blooms. It has also highlighted challenges that hinder our understanding of cyanobacterial global distribution and dominance including: failure to resolve withinspecies strain variability which reduces the predictive ability of deterministic models; failure to induce physiological characteristics of species in laboratory experiments, e.g., colony morphology in Microcystis which affects buoyancy regulation; introduced inaccuracy from inconsistent laboratory techniques across studies; failure to mimic realistic physical conditions such as light fluctuations and high light levels; and a variable degree of deviation between strains under culture conditions, etc. This thesis concludes with the need for laboratory studies that accurately reflect field conditions, including understanding of buoyancy, colony and interactions with turbulent mixing in formation of Microcystis blooms, to better align with global distribution of cyanoHABs.